LbfgsLogisticRegression(
this SweepableBinaryClassificationTrainers trainer,
string labelColumnName = "Label",
string featureColumnName = "Features",
SweepableOption <LbfgsLogisticRegressionBinaryTrainer.Options> optionBuilder = null,
LbfgsLogisticRegressionBinaryTrainer.Options defaultOption = null)
{
var context = trainer.Context;
if (optionBuilder == null)
{
optionBuilder = LbfgsLogisticRegressionBinaryTrainerSweepableOptions.Default;
}
optionBuilder.SetDefaultOption(defaultOption);
return(context.AutoML().CreateSweepableEstimator(
(context, option) =>
{
option.LabelColumnName = labelColumnName;
option.FeatureColumnName = featureColumnName;
return context.BinaryClassification.Trainers.LbfgsLogisticRegression(option);
},
optionBuilder,
new string[] { labelColumnName, featureColumnName },
new string[] { PredictedLabel },
nameof(LbfgsLogisticRegressionBinaryTrainer)));
}
private double Training(IDataView data, MLContext ctx)
{
var split = ctx.Data.TrainTestSplit(data, testFraction: 0.18);
var features = split.TrainSet.Schema
.Select(col => col.Name)
.Where(col => col != "Label")
.ToArray();
var trainer = new LbfgsLogisticRegressionBinaryTrainer.Options()
{
MaximumNumberOfIterations = 100,
};
var pipeline = ctx.Transforms.Concatenate("Features", features)
.Append(ctx.BinaryClassification.Trainers.Gam(learningRate: 0.052, numberOfIterations: 25000));
var model = pipeline.Fit(split.TrainSet);
var predictions = model.Transform(split.TestSet);
var metrics = ctx.BinaryClassification.Evaluate(predictions);
ctx.Model.Save(model, data.Schema, "model.zip");
_model = model;
return(metrics.Accuracy);
}
public override IEstimator <ITransformer> BuildFromOption(MLContext context, LbfgsOption param)
{
var option = new LbfgsLogisticRegressionBinaryTrainer.Options()
{
L1Regularization = param.L1Regularization,
L2Regularization = param.L2Regularization,
LabelColumnName = param.LabelColumnName,
FeatureColumnName = param.FeatureColumnName,
ExampleWeightColumnName = param.ExampleWeightColumnName,
NumberOfThreads = AutoMlUtils.GetNumberOfThreadFromEnvrionment(),
};
return(context.BinaryClassification.Trainers.LbfgsLogisticRegression(option));
}
public static void Example()
{
// Create a new context for ML.NET operations. It can be used for exception tracking and logging,
// as a catalog of available operations and as the source of randomness.
// Setting the seed to a fixed number in this example to make outputs deterministic.
var mlContext = new MLContext(seed: 0);
// Create a list of training data points.
var dataPoints = GenerateRandomDataPoints(1000);
// Convert the list of data points to an IDataView object, which is consumable by ML.NET API.
var trainingData = mlContext.Data.LoadFromEnumerable(dataPoints);
// Define trainer options.
var options = new LbfgsLogisticRegressionBinaryTrainer.Options()
{
MaximumNumberOfIterations = 100,
OptimizationTolerance = 1e-8f,
L2Regularization = 0.01f
};
// Define the trainer.
var pipeline = mlContext.BinaryClassification.Trainers.LbfgsLogisticRegression(options);
// Train the model.
var model = pipeline.Fit(trainingData);
// Create testing data. Use different random seed to make it different from training data.
var testData = mlContext.Data.LoadFromEnumerable(GenerateRandomDataPoints(500, seed: 123));
// Run the model on test data set.
var transformedTestData = model.Transform(testData);
// Convert IDataView object to a list.
var predictions = mlContext.Data.CreateEnumerable <Prediction>(transformedTestData, reuseRowObject: false).ToList();
// Print 5 predictions.
foreach (var p in predictions.Take(5))
{
Console.WriteLine($"Label: {p.Label}, Prediction: {p.PredictedLabel}");
}
// Expected output:
// Label: True, Prediction: True
// Label: False, Prediction: True
// Label: True, Prediction: True
// Label: True, Prediction: True
// Label: False, Prediction: False
// Evaluate the overall metrics.
var metrics = mlContext.BinaryClassification.Evaluate(transformedTestData);
PrintMetrics(metrics);
// Expected output:
// Accuracy: 0.87
// AUC: 0.96
// F1 Score: 0.87
// Negative Precision: 0.89
// Negative Recall: 0.87
// Positive Precision: 0.86
// Positive Recall: 0.88
// Log Loss: 0.37
// Log Loss Reduction: 0.63
// Entropy: 1.00
//
// TEST POSITIVE RATIO: 0.4760 (238.0/(238.0+262.0))
// Confusion table
// ||======================
// PREDICTED || positive | negative | Recall
// TRUTH ||======================
// positive || 210 | 28 | 0.8824
// negative || 35 | 227 | 0.8664
// ||======================
// Precision || 0.8571 | 0.8902 |
}
static void Main(string[] args)
{
MLContext ml = new MLContext(seed: 1);
var trainData = ml.Data.LoadFromTextFile <MlInput>("learningdata.csv", ',', hasHeader: true);
//var select = ml.Transforms.SelectColumns("Airlines", "Architecture", "Art & Museums", "Automotive", "Banking & Financial", "Cannabis", "Casinos & Lottery");
//var selectTransform = select.Fit(trainData).Transform(trainData);
var inputcolumns = new List <string> {
};
double f1sum = 0;
int count = 0;
var techlist = new List <string> {
"360 Photo/Video", "3D Environments", "3D Modelling", "AR", "AR Authoring", "AR Cloud", "AR Headsets", "AR Kit", "AR: Hololens", "AR: Magic Leap", "ARCore", "Avatar Creation", "Distribution & Device Management", "Eye Tracking", "Haptics & Peripherals", "Motion Capture",
"Motion Simulators", "Networking", "Photogrammetry", "Spatial Audio", "Unity", "Unreal", "VR", "VR Authoring", "VR Headsets", "VR: Mobile", "VR: PC", "Volumetric Capture", "Web XR"
};
foreach (var technology in techlist.ToList())
{
var removedlist = techlist.ToList();
removedlist.Remove(technology);
string[] removedarray = removedlist.ToArray();
var prepedData = ml.Transforms.DropColumns(removedarray);
var prepedDataTransform = prepedData.Fit(trainData).Transform(trainData);
TrainTestData trainTestData = ml.Data.TrainTestSplit(prepedDataTransform, testFraction: 0.2);
IEstimator <ITransformer> dataPipe = ml.Transforms.Concatenate("Features", new[] { "Airlines", "Architecture", "Art & Museums", "Automotive", "Banking & Financial", "Cannabis", "Casinos & Lottery", "Charities", "Education & Training", "Education (K-12)", "Emergency Response", "Health & Medical", "Industrial",
"Legal & Insurance", "Media & News & Entertainment", "Military", "Music", "Real Estate", "Restaurant & Food", "Retail", "Sales & Marketing", "Sports", "Telecommunications", "Travel & Tourism", "Collaboration & Social", "Communications", "Data Analytics", "Design", "Gaming", "General Training", "Health & Safety",
"LBE", "Leadership", "Marketing", "Rehabilitation", "Simulation", "Tools", "Training: Hands On", "Training: Hard Skills", "Training: Soft Skills" });
var options = new LbfgsLogisticRegressionBinaryTrainer.Options()
{
LabelColumnName = technology,
FeatureColumnName = "Features",
MaximumNumberOfIterations = 100,
OptimizationTolerance = 1e-8f
};
var lbfsgLogistic = ml.BinaryClassification.Trainers.LbfgsLogisticRegression(options);
var trainPipe = dataPipe.Append(lbfsgLogistic);
Display(trainTestData.TrainSet);
var model = trainPipe.Fit(trainTestData.TrainSet);
IDataView predictionView = model.Transform(trainTestData.TestSet);
CalibratedBinaryClassificationMetrics metrics = ml.BinaryClassification.Evaluate(predictionView, labelColumnName: technology);
//Console.WriteLine($"F1 {metrics.ConfusionMatrix.GetFormattedConfusionTable().ToString()} ");
Console.WriteLine($"F1 {metrics.F1Score} ");
if (metrics.F1Score > 0)
{
f1sum += metrics.F1Score;
count++;
}
}
Console.WriteLine($"Average: {f1sum/count}");
}
